Off-campus UMass Amherst users: To download campus access dissertations, please use the following link to log into our proxy server with your UMass Amherst user name and password.

Non-UMass Amherst users: Please talk to your librarian about requesting this dissertation through interlibrary loan.

Dissertations that have an embargo placed on them will not be available to anyone until the embargo expires.

Author ORCID Identifier


Open Access Dissertation

Document Type


Degree Name

Doctor of Philosophy (PhD)

Degree Program

Organismic and Evolutionary Biology

Year Degree Awarded


Month Degree Awarded


First Advisor

David King

Subject Categories

Population Biology | Terrestrial and Aquatic Ecology


Space use and movement patterns are integral to population dynamics and are often indicative of vulnerability to anthropogenic threats. Spatial ecology research can be fundamental to conservation strategies but is largely biased toward short-term intra- and interannual patterns. Without an understanding of space use over temporal scales commensurate with lifespan and the processes that may influence movement, conservation tools derived from short-term (2–4 yrs) movement patterns may be misguided or ineffective, particularly for long-lived species. The goal of Chapter 1 was to characterize the long-term (multi-decadal) spatial ecology of three long-lived (80–110 yrs) turtle species. We revisited six areas where the eastern box turtle (Terrapene carolina carolina), wood turtle (Glyptemys insculpta), and spotted turtle (Clemmys guttata) had been studied using telemetry 10–26 years prior with the goal of recapturing and tracking the same individuals that were previously studied in order to understand long-term patterns in space use. We tracked 60% of turtles (43 of 72) from original studies and observed generally high long-term fidelity, with overall 95% KDE home viii range overlap ranging approximately 40–60% across species. Nevertheless, long-term annual and overwintering fidelity was generally lower than short-term fidelity, highlighting that the temporal scale of typical telemetry studies does not completely characterize long-term movement. Moreover, we observed interspecific and intersexual differences in long-term fidelity, with spotted turtles — and male spotted turtles in particular — displaying comparatively low home range overlap at the multi-decadal scale. Females of all species generally displayed very high long-term seasonal fidelity, with no statistically- or biologically-significant difference between temporal scales for most seasons. In contrast, despite short-term fidelity similar to females, male spotted and wood turtles showed significantly lower long-term fidelity than short-term fidelity for nearly all seasons. These findings suggest that land protection directed toward known population activity areas is likely to remain effective for non-dispersing individuals for at least multiple decades. However, comprehensive land protection may require protecting suitable habitat and movement corridors within the surrounding landscape to capture long-term shifts in home range, particularly by males. Landscape context is integral to wildlife population ecology, affecting a range of life history parameters, yet very little is known about how landscape structure influences many taxa. In Chapter 2 we sampled wetlands across the eastern United States to examine the influence of landscape heterogeneity and anthropogenic land use on populations of freshwater turtles. Specifically, we aimed to understand how two components of landscape structure — compositional heterogeneity (wetland diversity) and configurational heterogeneity (wetland aggregation) — influence turtles with varying life history traits. Our results suggest that wetland configuration modulates the ix relationship between relative abundance of certain species and some anthropogenic landuse types. For example, spotted turtle (Clemmys guttata) and pond slider (Trachemys scripta) were negatively associated with human land use when wetlands were less aggregated, but this relationship subsided, and even became positive, as aggregation increased. These results demonstrate that some anthropogenic cover types are not strictly positive or negative for certain species, but are instead context-dependent. We also found that relative abundance generally increased (6 of 7 species) with higher wetland diversity, indicating that landscape supplementation may play an important role in turtle population ecology. We report a wide range of responses to roads that did not strictly correspond with well-established predictions related to body size and terrestrial activity patterns, including positive associations for certain species. This study supports the use of context driven approaches to land-use mitigation rather than blanket prescriptions. Climate change and land-use change are leading drivers of biodiversity decline, affecting demographic parameters that are important for population persistence. For example, scientists have speculated for decades that climate change may skew adult sex ratios in taxa that express temperature-dependent sex determination (TSD), but limited evidence exists that this phenomenon is occurring in natural settings. For species that are vulnerable to anthropogenic land-use practices, differential mortality among sexes may also skew sex ratios. In Chapter 3 we sampled the spotted turtle (Clemmys guttata), a freshwater species with TSD, across a large portion of its geographic range (Florida to Maine), to assess the environmental factors influencing adult sex ratios. We present evidence that suggests recent climate change may have skewed the adult sex ratio of spotted turtles, with predicted sex ratios following a pattern of increasing proportion of x females concomitant with warming trends, but only within the warmest areas sampled. At intermediate temperatures, there was no relationship with climate, while in the coolest areas we found the opposite pattern, with predicted sex ratios becoming more malebiased with increasing temperatures. These patterns might be explained in part by variation in relative adaptive capacity via phenotypic plasticity in nest site selection. Our findings also suggest that spotted turtles have a context-dependent and multi-scale relationship with land-use. We observed a negative relationship between male proportion and the amount of crop cover (within 300 m) when wetlands were spatially dispersed. However, when wetlands were aggregated, sex ratios remained consistent. This pattern may reflect sex-specific patterns in movement that render males more vulnerable to mortality from agricultural machinery and other threats. Our findings highlight the complexity of species’ responses to both climate change and land-use, and emphasize the role that landscape structure can play in shaping wildlife population demographics. Turtle populations are declining globally, yet limited attention has been directed toward understanding the conservation status of many species, particularly those perceived to be widespread and common. The goal of Chapter 4 was to contribute to the understanding of the conservation status of the eastern box turtle, a wide-ranging terrestrial generalist, in the northeastern United States (Maine to Virginia) by (1) characterizing relationships with anthropogenic land use and (2) estimating the extent of land-use driven habitat impairment for the region. We used a regional dataset of occurrence records combined with pseudo-absences to develop species distribution models to first estimate the potential distribution in the northeastern U.S. and then predict habitat suitability within that distribution. We observed a strong positive relationship xi between probability of occurrence and canopy cover (within 180 m) and a strong negative relationship with hay/pasture fields (360 m), cultivated crops (180 m), imperviousness (360 m), and forest loss primarily from timber harvesting (since 2000; 1,440 m). We estimate that approximately 51% of eastern box turtle habitat in the northeastern U.S. is impaired by land use. While our results indicate impairment throughout the region, the majority of habitat loss is predicted from Pennsylvania and Delaware to Virginia. This study, in combination with previous long-term studies documenting population declines, provides evidence of possible widespread population decline, and suggests that greater attention to the conservation status of the eastern box turtle is warranted, particularly within the northeastern U.S. While efforts to understand distributional responses to climate change typically occur at the species level, a growing body of research highlights the importance of considering intraspecific variation in responses to climate when forecasting future conditions. Without experimental evidence demonstrating local adaptation, studies typically use information such as geographic isolation or genetic affinity to subdivide a global population before modeling and projecting subunits under future conditions. However, local adaptation can occur in the absence of geographic isolation, and rangewide genetic studies are costly and often unavailable; therefore, alternative methods for species subdivision are needed. In Chapter 5, we explore the utility of using climate itself to guide subdivision, whereby population groupings are determined by patterns of distinctive climate structure. Using the globally threatened spotted turtle as a focal species, we employed multivariate clustering and principal component analyses to identify three groups of occurrences that experience distinctive climate conditions. xii Notably, the most distinctive population grouping did not correspond with a geographically isolated portion of the range and therefore would have been ignored if geographic isolation was used as the only subdivision criteria. We developed four distribution models: one for each of the three intraspecific subunits identified, and one for the global population. When comparing single-model and local population-based predictions, we observed a reduction in niche overlap of approximately 43–50% under future scenarios, suggesting that modeling the distribution as a single entity could lead to inaccurate forecasts if these population subunits are biologically relevant. Both modeling approaches predicted substantial losses (>50%) of currently suitable habitat under future scenarios, but overall, local population models predicted more severe losses than the single model. Both modeling approaches suggested that the midwestern portion of the species range will remain more stable, proportionally retaining more currently suitable habitat compared to the coastal portion of the range. In the absence of rangewide genetic data, climate structure offered a useful means by which to identify intraspecific subunits that would not have been observed if only geographic isolation were used.


Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Available for download on Sunday, May 26, 2024